it has been 12 hours since openai announced o1 and it has so far failed to to solve
- Riemann hypothesis
- Quantum Gravity
- FTL (Faster Than Light travel)
- P=NP
- Grand Unified Theory
- Cure for cancer
clearly this shows ai has hit a wall and openai is about to go bankrupt
@harambe_musk
seems like shrek entropy sampler with early exit solves if not reduce the hallucination problem significantly with big boy models. Some people are running evals now so far seems promising
@ricepaddyhacker
qr code at seat/ticket use the app to scan it to register the positions and sync with controller the stick talks to app over blue tooth
@ricepaddyhacker
qr code at seat/ticket use the app to scan it to register the positions and sync with controller the stick talks to app over blue tooth
Once again, an AI system is not "thinking", it's "processing", "running predictions",... just like Google or computers do.
Giving the false impression that technology systems are human is just cheap snake oil and marketing to fool you into thinking it's more clever than it is.
You all want to know why apple is so far behind with regards to AI/ML ? The ethics crowd basically destroyed the engineering teams and this paper is a legacy of that
LLMs don’t reason. But that’s OK, neither do we. What we imagine to be “reasoning” is an abstraction that has never been realized in any strict sense in any stochastic natural phenomenon, such as human brain.
funny thing about the current state of the space is that it's pretty evident that everyone is independently discovering the same thing and tricks and architecture but due to a multitude of reasons it cannot be directly talked about or shared
Hearing that AI models are collapsing because AI bros poisoned their own well by flooding the internet with loads of slop that’s being fed back into the training data with no way to sift it out. It fills me with such schafenfreude
Our latest work AlphaProof, building on AlphaZero, LLMs and the
@leanprover
theorem prover, combined with AlphaGeometry 2 managed to solve 4 IMO problems and achieve silver-medalist level! 🚀
More at
GPT-4o, o1, AVM—marvels, but like AI Dungeon in 2021, just better. The core limit: reactive, not proactive. They won’t hit you up or activate themselves. No agency. Idle unless asked. Frozen, no time, locked in one moment.
I hope that changes.
The NYT got ahold of docs from the OpenAI funding round.
- 350 million people used Chat in Aug
- huge user growth after anon login
- 10 mill active subscribers
- sub going up to $2 by the end of the year, the sub will scale up to $44 by 2029
- projected yearly rev by 2029: $100B
@ricepaddyhacker
qr code at seat/ticket use the app to scan it to register the positions and sync with controller the stick talks to app over blue tooth
It seem our friendly Tech shrek completed another orbit around the sun recently.
Happy birthday
@_xjdr
hope you had a good one cant believe you were dropping ground breaking research and merging prs on your birthday
Holy shit! If this is real this is huge.
JWST has spotted a massive non-human object heading towards us from 10 light years away that also seems to be changing its trajectory.
This is what the Congress was apparently briefed about in a classified briefing.
NEW: OpenAI now has a $4B revolving line of credit, bringing its total liquidity to more than $10B. Based on current rates, OpenAI will be paying roughly 6% on money that it borrows right away.
Participating banks: JPMorgan Chase, Citi, GS, MS & more.
Ok so the code for multiple branching strategy is working somewhat it looks ahead 10 tokens and previous 10 tokens selects a branching strategy dynamically based on entropy ratio, attention weights and perplexity
caveats
1) this has NOTHING to do with how my employer is doing
@yacineMTB
not a big enough of a concern actually and you dont want to risk a dirty bomb in case the explosion on the firy end decides to com up towards the pointy end
It has been 48 hours since OpenAI released the o1 model, and mental gymnastics and goal post moving is off the charts.
10 wild example of Olympic level goal post moving 🧵
OAI is all hype but have let me check my notes solved reasoning and deployed to millions of users. While let me check my notes again, no other lab has come close to doing
Guys, Strawberry is literally just baked in Chain-of-Thought reasoning which came out in January of 2022. OpenAI has nothing. It's all hype.
Original CoT paper was published at NeurIPs Jan 2022.
dont get me wrong this level of tele-operated bots are game changing. it will unlock so much but its also kinda dick move not to explicitly state that they are
@basedjensen
Entropy handling mitigates hallucination. Hallucination occurs when a model is uncertain but is compelled to output an answer token and continue its answer anyway. Entropy/uncertainty offers a chance to self-correct.
added reflection backtracking and reward mechanics and tags to the code and messed around with the prompt a bit. I think we have a decent reasoner at home now
Ok so the code for multiple branching strategy is working somewhat it looks ahead 10 tokens and previous 10 tokens selects a branching strategy dynamically based on entropy ratio, attention weights and perplexity
caveats
1) this has NOTHING to do with how my employer is doing
@yacineMTB
90% of S tier conference papers in ML just put fancy math for the sake of it. At msft research we made a distillation algorithm and then wrote 1 page full of theory after we had all the numbers just for the sake of it 😭 we thought of the theory 1 week before deadline 💀
.
@WorldLibertyFi
Token Sale goes live on Tuesday morning, October 15th! This is YOUR chance to help shape the future of finance. Be there on Monday, October 14th at 8 AM EST for an Exclusive Spaces to learn more. Join the whitelist today and be ready for Tuesday:
MF you all telling me rope is this simple ? why you all not naming your variables properly so rest of us can read them.... JFC i thought some magic math bs was happening when i looked at the code previously
i dont have the audience to reach outside the bubble but here is the thing... synthetic data flywheel is so good right now there is no realistically possible way to poison the dataset in any way. Anyone who claims they can is pushing a grift like night shade
big day for openai: chatgpt is generating more than $225 million in revenue per month (and that's a conservative estimate) based on new usage metrics we've gotten
@thirk
actually we tested 100 dev devices found out it was terrible I got sloshed that night and rewrote the whole stack overnight and tested it next day with batches of 100 200 I think max we did was 250
🎥 Today we’re premiering Meta Movie Gen: the most advanced media foundation models to-date.
Developed by AI research teams at Meta, Movie Gen delivers state-of-the-art results across a range of capabilities. We’re excited for the potential of this line of research to usher in
me: babe Demis and google deep mind got Nobel for chemistry for making ai for Proteins
her: so when are you getting yours? i have perfect outfit picked out for when you get yours
....